the response of clamped sandwich plates with metallic · pdf filethe response of clamped...
TRANSCRIPT
The response of clamped sandwich plates with metallic foam cores to simulated blast
loading
D.D. Radford, G.J. McShane, V.S. Deshpande and N.A. Fleck∗
Department of Engineering, University of Cambridge
Trumpington St., Cambridge, CB2 1PZ, UK
Abstract
The dynamic responses of clamped circular monolithic and sandwich plates of equal areal
mass have been measured by loading the plates at mid-span with metal foam projectiles. The
sandwich plates comprise AISI 304 stainless steel face sheets and aluminium alloy metal
foam cores. The resistance to shock loading is quantified by the permanent transverse
deflection at mid-span of the plates as a function of projectile momentum. It is found that the
sandwich plates have a higher shock resistance than monolithic plates of equal mass. Further,
the shock resistance of the sandwich plates increases with increasing thickness of sandwich
core. Finite element simulations of these experiments are in good agreement with the
experimental measurements and demonstrate that the strain rate sensitivity of AISI 304
stainless steel plays a significant role in increasing the shock resistance of the monolithic and
sandwich plates. Finally, the finite element simulations were employed to determine the
pressure versus time history exerted by the foam projectiles on the plates. It was found that
the pressure transient was reasonably independent of the dynamic impedance of the plate,
suggesting that the metal foam projectile is a convenient experimental tool for ranking the
shock resistance of competing structures.
Keywords: impact loading, sandwich plates, metallic foams, finite elements
Revised version submitted to Int. J. Solids Structures, June 2005
∗ Corresponding author, Tel.: +44-1223 332650; fax: +44-1223 765046 Email address: [email protected]
1
1. Introduction
Clamped monolithic plates are commonly used in the design of commercial and military
vehicles. For example, the outermost hull of a ship comprises plates welded to an array of
stiffeners. The superior performance of sandwich plates relative to monolithic solid plates
is well established for applications requiring high quasi-static strength. However, the
resistance of sandwich plates to dynamic loads remains to be fully investigated in order to
quantify the advantages of sandwich design over monolithic design for potential
application in shock resistant structures.
Over the past fifty years, the response of monolithic beams and plates to shock type
loading has been extensively investigated. For example, Wang and Hopkins (1954) and
Symmonds (1954) have analysed the impulse response of clamped circular monolithic
plates and clamped monolithic beams, respectively. Their analyses were restricted to
infinitesimal strains and displacements. Jones (1968) presented an approximate analytical
solution for the finite deflection of simply supported circular monolithic plates by direct
application of the principle of virtual work for an assumed deformation mode. More
recently, Fleck and Deshpande (2004) have proposed an analytical model for the finite
deflection response of clamped sandwich beams subjected to shock loadings, including the
effects of fluid-structure interaction. They have demonstrated the accuracy of their
analytical model in the case of no fluid-structure interaction by direct comparison with the
finite element calculations of Xue and Hutchinson (2004) for clamped sandwich beams.
The analytical model of Fleck-Deshpande has been extended by Qiu et al. (2005) for
clamped circular sandwich plates, and is again supported by finite element calculations of
Xue and Hutchinson (2003) for sandwich plates with a foam-like core and the effects of
fluid-structure interaction again neglected. These recent investigations have each
demonstrated that circular sandwich plates can posses a superior resistance to shock
loading than monolithic plates of the same areal mass.
To date, there have been little experimental data to support or refute these predictions, as it
is difficult to perform laboratory scale experiments with a prescribed dynamic loading
history. In the recent study by Radford et al. (2005a), an experimental technique was
introduced to subject structures to high intensity pressure pulses using metal foam
2
projectiles. The applied pressure versus time pulse attempts to mimic shock loading in air
and in water, with peak pressures on the order of 100 MPa and loading times of
approximately 100 µ s. This experimental technique has been employed by Radford et
al. (2005b) and Rathbun et al. (2004) to investigate the dynamic response of clamped
sandwich beams with metal foam and lattice cores. These studies indicated that the
sandwich beams have a superior shock resistance compared to monolithic beams of the
same mass. However, no attempt was made to de-convolute the cross-coupling between
the dynamic response of the beams and that of the projectile, making it difficult to quantify
the enhanced shock resistance of the sandwich beams.
In this study, we employ the experimental technique of Radford et al. (2005a) to explore
the shock resistance of clamped circular monolithic and aluminium foam core sandwich
plates of equal mass. First, the manufacturing route of the sandwich plates is detailed and
the experimental procedure is described for loading the plates at mid-span by metal foam
projectiles. The experimental results are compared with finite element predictions. Finally,
the finite element simulations are used to determine the cross-coupling between the foam
projectile loading and the plate response.
2. Experimental investigation
In this study, metal foam projectiles are used to load dynamically clamped circular
sandwich plates comprising an aluminium metal foam core and AISI 304 stainless steel
face sheets. The main aims of the experimental investigation are:
(i) To compare the dynamic resistance of sandwich plates with monolithic plates
(made from the sandwich plate face sheet material) of equal mass.
(ii) To investigate the effect of the sandwich core thickness upon the shock resistance.
(iii) To demonstrate the accuracy of finite element predictions for the dynamic
response.
3
2.1 Specimen configuration and manufacture
Circular sandwich plates comprising Alporas1 aluminium metal foam cores and AISI type
304 stainless steel face sheets were manufactured to a net areal mass m of 23.5 kgm-2.
The plates, of radius 80R = mm, comprised two identical AISI 304 stainless steel face
sheets, of thickness h and density 8060fρ = kgm-3, and an Alporas aluminium foam
core, of density 430cρ ≈ kgm-3 and thickness c , see Fig. 1. In order to achieve a common
areal mass of cf chm ρρ += 2 = 23.5 kgm-2, face sheets of thickness 1 18h .= mm and
0.88 mm were employed for cores of thickness 10c = mm and 22 mm, respectively.
The sandwich plates were manufactured as follows. Alporas foam cores of rectilinear
dimension 250 mm 250 mm c× × were electro-discharged machined from blocks of foam.
Stainless steel face sheets of dimensions 250 mm 250 mm h× × were degreased and
abraded, and were then adhered to the foam core using Redux 322 epoxy adhesive on a
nylon carrier mesh. The sandwich plates were air-cured at 175oC for 1.5 hours, and
bonding was facilitated by imposing dead-loading with a nominal contact pressure of
0.01 MPa. Finally, eight equally spaced clearance holes for M8 bolts were drilled in these
plates on a pitch circle of radius 102 mm.
The clamped plate geometry is sketched in Fig. 2. Each sandwich plate was clamped
between two annular steel rings of thickness 7 mm, inner radius 80 mm and outer radius
125 mm, on a pitch circle of radius 102 mm. The assembly was then bolted down onto a
rigid loading frame by M8 bolts, as sketched in Fig. 2.
In addition to the dynamic tests on two configurations of sandwich plates, dynamic tests
were performed on AISI 304 stainless steel monolithic circular plates of areal mass
m=23.5 kgm-2 for comparison purposes. These circular monolithic plates of radius
80R = mm and thickness 2 92h .= mm were gripped using the same apparatus as that
described above.
1 Shinko Wire Co. Ltd., Amagasaki, Japan.
4
2.2 Properties of the constituent materials
The uniaxial compressive responses of the Alporas metal foam was measured quasi-
statically at a nominal strain rate of 0.001 s-1 using two sizes of cuboidal specimens, with a
side-length 10L = mm and 22 mm. These lengths correspond to the two core thicknesses
employed in the sandwich tests. The measured compressive nominal stress versus nominal
strain responses are plotted in Fig. 3a. The overall responses of the two cores are similar,
with a compressive plateau strength of 3.5 MPa and a nominal lock-up strain Dε of 0.8.
The uniaxial tensile responses of the three thicknesses of AISI 304 stainless steel used in
the sandwich and monolithic plates were measured at a nominal strain rate of 0.001 s-1,
and the true tensile stress versus logarithmic strain curves are plotted in Fig. 3b. The
tensile responses of the three thicknesses of AISI 304 stainless steel are qualitatively
similar with a 0.2% offset yield strength of 300 MPa, followed by a linear hardening
response up to an ultimate tensile strength of 1150 MPa.
The sandwich panels made from the Alporas metal foam and AISI 304 stainless steel face
sheets were tested under dynamic loading and thus a knowledge of the high strain rate
behaviour of the metal foams and AISI 304 stainless steel is necessary for accurate finite
element simulations of the experiments. While the Alporas metal foam has a small strain
rate sensitivity (Dannemann and Lankford, 2000 and Miyoshi et al., 2002), shock wave
propagation in these foams becomes important at impact velocities exceeding about
50 ms-1 (Radford et al. 2005a). The algorithm employed in the finite element simulations
to capture the strain rate and shock wave effects are given in Section 4, with additional
details given in Appendix A.
Stout and Follansbee (1986) have investigated the strain rate sensitivity of AISI 304
stainless steel for strain rates in the range 4 -1 4 -110 s 10 sε− < <& by performing a series of
compression tests. In order to present their results, we introduce the strength enhancement
ratio R as the ratio of the stress ( 0 1)pd .σ ε = at any applied strain rate pε& to the reference
stress 0 ( 0 1)p .σ ε = at an applied 3 -110 spε −=& . Their data for AISI 304 stainless steel is
re-plotted in Fig. 4 in the form of R versus the plastic strain rate pε& for 1414 1010 −−− << ss pε& . The strength enhancement ratio R is reasonably independent of the
5
choice of the plastic strain pε at which R is calculated. Thus, the dynamic strength dσ
versus plastic strain pε history can be expressed as
0( )= ( ) ( ) p p pd Rσ ε ε σ ε& , (1)
where ( )pR ε& is given in Fig. 4a. In the finite element simulations of the dynamic response
of the sandwich panels presented in Section 4, we shall employ this prescription for the
strain rate sensitivity of the AISI 304 stainless steel, with 0 ( )pσ ε given by the measured
quasi-static stress versus strain histories for the various thicknesses of AISI 304 stainless
steel (Fig. 3b). As an example, the estimated true tensile stress versus logarithmic plastic
strain histories for the 1.18 mm thick AISI 304 stainless steel at four selected values of
applied strain rate are sketched in Fig. 4b.
2.3 Test protocol
Alporas aluminium foam projectiles were used to impact the clamped monolithic and
sandwich plates over a central circular patch of radius a, as shown in Fig. 1. The use of
foam projectiles as a means of providing well-characterised pressure versus time has
recently been developed by Radford et al. (2005a) and then subsequently employed by
Radford et al. (2005b) to investigate the dynamic response of clamped sandwich beams
with metal foam cores and with lattice cores.
Circular cylindrical projectiles of length ≈0l 50 mm and diameter 28 5d .= mm were
electro-discharge machined from Alporas foam blocks of density pρ in the range
380 kgm-3 to 490 kgm-3. The projectiles were fired from a gas gun of barrel diameter
28.5 mm and length 4.5 m. The projectile velocities 0v ranged from 160 ms-1 to 570 ms-1,
providing a projectile momentum per unit area 000 vlI pρ= of up to 13 kNs m-2. Table 1
summarises the set of experiments performed, with details of the projectile density, length,
impact velocity and initial momentum. The plates were examined after each test to
measure the permanent deflection of the faces, and to detect any visible signs of failure.
3. Experimental results
6
For each specimen configuration, at least four levels of initial momentum were applied by
varying the density, length and impact velocity of the foam projectiles. Table 1
summarises the observed permanent back-face transverse deflection at the mid-span and
the amount of core compression for each specimen.
The effect of core thickness upon the shock resistance is summarised in Fig. 5. It contains
a plot of the permanent back-face deflection at mid-span versus the momentum of the
foam projectile 0I . The plate of core thickness 22c = mm deflects less than the sandwich
plate of core thickness 10c = mm over the entire range of 0I investigated. The mid-span
deflections of the monolithic plates of equal areal mass are included in Fig. 5. Intriguingly,
the monolithic plate deflects less than the sandwich plates at low values of 0I .
The final core compressive strain is defined as ccc /∆=ε , where c∆ is the reduction in
core thickness at mid-span; the measured dependence of the cε upon 0I is given in Fig. 6
for the two sandwich plate configurations. The core compressive strain increases with
increasing 0I , and is consistently less for the thicker core. At the highest impulse
( 0 13I ≈ kNs m-2), both sandwich plates have fully compressed such that the lock-up strain
of the metal foam core has been achieved at mid-span.
In order to gain insight into the dynamic deformation mechanisms, the monolithic and
sandwich plates tested at 0 13I ≈ kNs m-2 were sectioned along their diametral planes.
Photographs of the diametral sections are shown in Fig. 7a for the monolithic plate
(specimen M9, back-face deflection of 14.2 mm), in Fig.7b for the sandwich plate of core
thickness 10 mm (specimen A4, back-face deflection of 12.8 mm) and in Fig.7c for the
sandwich plate of core thickness 22 mm (specimen B4, back-face deflection of 9.2 mm).
The diametral profiles show that significant plastic deformation occurred in the vicinity of
the foam impact and that the plates are continuously curved. This suggests that the
dynamic deformation of plates involves the formation of travelling hinges, analogous to
the behaviour of monolithic and sandwich beams (Radford et al. 2005b). Delamination of
the core from the face sheets is evident in Fig. 7c; it occurred after the specimen was
sectioned, rather than during the impact experiment.
7
For comparison purposes, a clamped monolithic plate and the two configurations of
sandwich plates were loaded quasi-statically over a central patch. These experiments were
performed at a displacement rate of 510− ms-1 using a flat-bottomed steel cylindrical
punch, of diameter 28.5 mm. The specimens were loaded until the back-face deflection at
mid-span matched that in the dynamic experiments (Fig. 7). Upon unloading the
specimens were sectioned along their diametral planes. The photographs shown in Fig. 8
reveal membrane action, with stationary plastic hinges at the periphery of the indenter and
at the supports (as evidenced by the discontinuity in inclination of the plates at those
locations).
4. Finite element simulations
Comparisons of the finite element (FE) simulations and the measured responses of the
monolithic and sandwich plates are presented in this section. All computations were
performed using the explicit time integration version of the commercially available finite
element code ABAQUS2 (version 6.4). The circular plates were modelled using four noded
axisymmetric quadrilateral elements with reduced integration (CAX4R in ABAQUS
notation). Clamped boundary conditions, with vanishing displacements in the radial and
tangential directions, were prescribed on the outer radius of the plate, r R= . Dynamic
loading of each plate was simulated by impact of a foam projectile, as described
subsequently. Typically, there were 160 elements along the radius of each plate and
approximately three elements per millimetre through the thickness of the face sheets. The
prescription employed to set the mesh density in the foam core (and foam projectile) is
given in Appendix A. Mesh sensitivity studies revealed that further refinements did not
appreciably improve the accuracy of the calculations.
The cylindrical foam projectiles of diameter 28.5 mm were also modelled using CAX4R
axisymmetric elements, with contact between the outer surface of the projectile and the top
surface of the plates modelled by a frictionless contact surface as provided by ABAQUS.
At the start of the simulation, the projectile was imparted with a uniform initial velocity ov
and was brought into contact with the plate at its mid-span.
8
4.1 Constitutive description
The AISI 304 stainless steel face sheets of the sandwich plates were modelled by J2-flow
theory rate dependent solids of density 8060fρ = kgm-3, Young’s modulus 210E = GPa
and Poisson ratio 0 3.ν = . The uniaxial tensile true stress versus equivalent plastic strain
curves at plastic strain rates 10-3 s-1 ≤≤ pε& 104 s-1 were tabulated in ABAQUS using the
prescription described in Section 2.2. Reference calculations were also performed, with the
rate dependence of the AISI 304 stainless steel neglected by setting ( )pR ε& =1 in equation
(1). Thus, these rate independent calculations made direct use of the quasi-static tensile
stress versus strain response of the AISI 304 stainless steels (Fig. 3b).
The foam core and the projectile were modelled as a compressible continuum using the
metal foam constitutive model of Deshpande and Fleck (2000). Write ijs as the usual
deviatoric stress and the von Mises effective stress as 3 2e ij ijs s /σ ≡ . The isotropic yield
surface for the metal foam is specified by
0ˆ Y ,σ − = (2)
where the equivalent stress σ̂ is a homogeneous function of eσ and mean stress
3m kk /σ σ≡ according to
( )
2 2 2 22
11 3
e mˆ/
σ σ α σα
⎡ ⎤≡ +⎣ ⎦+. (3)
The material parameter α denotes the ratio of deviatoric strength to hydrostatic strength,
and the normalisation factor on the right hand side of relation (3) is chosen such that σ̂
denotes the stress in a uniaxial tension of compression test. An over-stress model is
employed with the yield stress Y specified by
pcˆY ηε σ= +& , (4)
2 Hibbit, Karlsson and Sorensen Inc.
9
where η is the viscosity, pε̂& the plastic strain rate (work conjugate to σ̂ ), and p( )c ˆσ ε is
the static uniaxial stress versus plastic strain relation. Normality of plastic flow is
assumed, and this implies that the “plastic Poisson’s ratio” 22 11p p
p /ν ε ε= − & & for uniaxial
compression in the 1-direction is given by
( )( )
2
2
1 2 3
1 3p
/ /
/
αν
α
−=
+ (5)
In the simulations, the Alporas foam is assumed to have a Young’s modulus 1 0cE .= GPa,
an elastic Poisson’s ratio 0 3.ν = and a plastic Poisson’s ratio 0pν = (Ashby et al. 2000).
The static yield strength cσ versus equivalent plastic strain pε̂ history is calibrated using
the compressive stress versus strain responses presented in Fig. 3a. Data from the 22 mm
thick foam specimen is used for the foam cores of thickness 22=c mm and for the
projectiles, while data from the 10 mm thick specimen is employed to model the foam core
of thickness 10=c mm. A prescription for the viscosity η is given in Appendix A.
The geometries of the simulated monolithic and sandwich plates match those of the
experiments (cf. Section 2.1), with the density of the sandwich cores taken to be
432cρ = kgm-3 and 428 kgm-3, for the sandwich plates of core thickness 22=c mm and
10 mm, respectively. The density and length of the foam projectiles are as listed in
Table 1.
4.2 Comparison of finite element predictions and measurements
Unless otherwise stated, all calculations reported in this study have been performed with
the rate dependent constitutive response for the AISI 304 stainless steel. Sample FE
predictions of the mid-span deflection versus time histories of the monolithic plate
specimen M9 and the sandwich plate specimen B4 ( 22c = mm) are plotted in Fig. 9, with
0I =13 kN s m-2 in both cases. The figure shows the mid-span deflection of the back-face
of the monolithic plate and of both front and back faces of the sandwich plate. The
deflection time histories indicate that only small elastic vibrations occur after the peak
10
deflection has been attained. Moreover, the peak deflection is approximately equal to the
final permanent deflection for both the monolithic and sandwich plates.
The predicted and measured mid-span, back-face deflections of the monolithic plates are
shown in Fig. 10. The predictions are given for both the rate dependent and rate
independent constitutive descriptions of the AISI 304 stainless steel. The permanent
deflections in the FE calculations are estimated by averaging the displacements near the
end of the calculations (over the time interval =t 1.5 ms to 3.0 ms). It is concluded that
the rate dependent FE model predicts the permanent deflection to reasonable accuracy,
with a slight over-prediction at high values of 0I . On the other hand, the rate independent
model substantially over-predicts the experimental measurements, especially for high 0I .
Similar comparisons of the predicted and measured permanent mid-span deflections are
shown in Fig. 11 for the back-face of the sandwich plates. Again, good agreement between
the rate dependent FE predictions and the measurements is seen. The FE predictions of the
final core compression cε have been included in Fig. 6; these predictions assume the rate
dependent model for the AISI 304 face sheets. The FE simulations predict the
measurements with reasonable accuracy, but give a slight over-prediction of cε at low 0I .
4.3 Pressure versus time histories exerted by foam projectiles
The pressure versus time history exerted by the foam projectile on a structure depends
upon the foam projectile (i) density pρ , (ii) length ol , (iii) compressive stress versus
strain response of the foam, and (iv) projectile velocity ov (Radford et al. 2005a). The
analysis of Radford et al. (2005a) suggests that metal foam projectiles exert a rectangular
pressure versus time pulse of magnitude
D
pc
vp
ερ
σ20
0 += , (6)
and duration
11
o D
o
lvετ = , (7)
on a rigid stationary target. Here, cσ and Dε are the plateau stress and nominal
densification strain of the foam, respectively. This pressure versus time history, however,
is sensitive to the structural response of the target: a decrease in the mechanical impedance
of the target leads to a decrease in the applied pressure and to an increase in the pulse
duration. Thus, it is unclear whether the observed differences in response of the various
monolithic and sandwich plates for a given 0I are due to differences in their intrinsic
impact resistance, or a result of the application of different pressure time histories on the
plates. We shall employ the FE calculations to clarify this ambiguity.
The pressure versus time history exerted by the foam projectile on the plates is derived
from the FE simulations as
)()( 0 tvltp p&ρ= , (8)
where v( t ) is the average axial velocity of the foam projectile at time t , and the over-dot
denotes time differentiation. Momentum conservation implies that,
0
oI p( t )dt∞
= ∫ . (9)
The accuracy of the pressure versus time histories, as extracted from the FE calculations
using equation (8), was confirmed by performing additional FE calculations on the
monolithic and sandwich plates with the pressure transients taken as the loading. Both the
rate dependent and rate independent versions of the constitutive law for the AISI 304 face
sheets were employed. In all cases, the deflections were within 0.5 % of those obtained
previously using the projectile loading.
The calculated pressure versus time histories exerted by the foam projectiles on the
monolithic plate M9, the sandwich plate A4 ( 10c = mm) and on the sandwich plate B4
12
( 22c = mm) are plotted in Fig. 12; in each case 0I =13 kN s m-2. (All FE calculations
detailed here and subsequently have been performed using the strain rate sensitive model
for the AISI 304 stainless steel.) While the differences between the pressure versus time
histories are small, it is clear from Fig. 12 that the pressure pulse duration is shortest for
the monolithic plate. Consequently, the average pressure exerted by the foam projectile on
the monolithic plate is highest.
We quantify the effect of these differences in the pressure versus time histories upon the
structural response of the plates in three steps. First, the pressure versus time histories
exerted on the sandwich plates by the foam projectiles are extracted from the FE
simulations using the above prescription. Second, the pressure versus time histories are
applied to the monolithic plate on a central patch of diameter 28 5d .= mm, and the
resulting mid-span deflection versus time history is recorded for a period of 3 ms. And
third, the permanent mid-span deflection of the monolithic plate is evaluated from these
deflection versus time histories.
FE predictions of the permanent back-face deflections at mid-span of the 10c = mm
sandwich plate and of the monolithic plate are plotted in Fig. 13a, each for the case of
direct loading by the foam projectiles. Additionally, FE predictions are given for the mid-
span deflection of the monolithic plates due to the pressure versus time loading history as
exerted by the foam projectile on the sandwich plate; these predictions are labelled
‘pressure loading’. Imposition of the pressure loading results in a slight reduction in the
mid-span deflections of the monolithic plates, especially at large 0I . However, the overall
conclusion that sandwich plates outperform monolithic plates at high values of 0I remains
unchanged. Similar comparisons of the FE predictions of the permanent back-face
deflections at mid-span of the 22c = mm sandwich plates and monolithic plates due to
loading by foam projectiles are shown in Fig. 13b. Included in the figure are the associated
permanent mid-span deflections of the monolithic plates loaded by the pressure versus
time histories exerted by the foam projectile on the 22c = mm sandwich plates. Again, the
pressure loading results in a small reduction in the mid-span deflections of the monolithic
plates at high values of 0I , but the overall conclusion that the 22c = mm sandwich plates
outperform the monolithic plates remains unchanged. We conclude that foam projectiles
13
are a convenient laboratory tool for investigating the dynamic response of such monolithic
and sandwich plates.
5. Concluding remarks
Metal foam projectiles have been used to impact clamped circular plates of monolithic and
sandwich construction, with a metal foam core. The permanent deflections, and core
compression of the sandwich plates, have been measured as a function of the projectile
momentum and the measured response are compared with finite element simulations. It is
found that the deformation mode due to dynamic loading is significantly different to that
observed in quasi-static loading, due in part to the occurrence of travelling plastic hinges
in the dynamic case. The sandwich plates outperform monolithic plates of equal mass at
sufficiently high values of projectile momentum. An increase in the core thickness (while
keeping the areal mass fixed) enhances the shock resistance of the sandwich plates.
The finite element simulations capture the observed response with reasonable accuracy,
and reveal that the strain rate sensitivity of the AISI 304 stainless steel plays a significant
role in increasing the shock resistance of the monolithic and sandwich plates. It is shown
explicitly that the pressure versus time history imparted on the clamped monolithic plates
is similar to that imparted on the sandwich plates. Consequently, the structural response of
the clamped plates is dictated mainly by the momentum of the foam projectile. The metal
foam projectile is thereby a useful laboratory tool for exploring the shock resistance of
monolithic and sandwich structures.
Acknowledgements
The authors are grateful to ONR for their financial support through US-ONR IFO grant
number N00014-03-1-0283 on The Science and Design of Blast Resistant Sandwich
Structures and to the Isaac Newton Trust, Trinity College Cambridge.
14
Appendix A: Prescription of the metal foam viscosity in the FE calculations
The prescription for choosing the viscosity η in the rate dependent metal foam
constitutive law and the associated choice of the finite element mesh size e employed in
the foam core and projectile are discussed in this Appendix.
Dynamic compression of a rate dependent foam gives rise to a shock wave of finite width
as discussed in (Radford et al. 2005a). For a linear viscous foam, the shock width w is
given by (Radford et al. 2005a)
Dwv
ηερ
=∆
, (A1)
where η is the linear viscosity, ρ the initial foam density, Dε the nominal densification
strain of the foam, and v∆ the velocity jump across the shock. For the purposes of this
discussion v∆ is approximately equal to the projectile velocity ov . In the finite element
calculations, we choose η such that the shock width w is much less than the core
thickness c of the sandwich and less than the projectile length ol . This ensures that the
viscosity η does not significantly affect the structural response.
Large gradients in stress and strain occur over the shock width w . To ensure that the finite
element calculations resolve these large gradients accurately, the element size e in the
core was taken to satisfy the condition 10/we ≤ .
15
References
Ashby, M. F., Evans, A. G., Fleck, N.F., Gibson, L.J., Hutchinson, J.W., Wadley, H. N.
G., 2000. Metal Foams: A Design Guide. Butterworth-Heinemann, Oxford.
Dannemann, K. A., Lankford, J., 2000. High strain rate compression of closed-cell
aluminium foams. Materials Science and Engineering: A 293(1-2): 157-164.
Deshpande, V. S., Fleck, N.A., 2000. Isotropic constitutive models for metallic foams.
Journal of Mechanics and Physics of Solids 48: 1253-1283.
Fleck, N. A. Deshpande, V.S., 2004. The resistance of clamped sandwich beams to shock
loading. Journal of Applied Mechanics, ASME 71(3): 386-401.
Jones, N., 1968. Impulsive loading of a simply supported circular rigid plate. Journal of
Applied Mechanics 27: 59.
Miyoshi, T., Mukai, T., Higashi, K., 2002. Energy absorption in closed-cell Al-Zn-Mg-Ca-
Ti foam. Materials Transactions 43(7): 1778-1781.
Qiu, X., Deshpande, V.S., Fleck, N.F., 2005. Impulsive loading over a central portion of
clamped sandwich beams. Journal of Mechanics and Physics of Solids (in press).
Radford, D. D., Deshpande, V.S., Fleck, N.A., 2005a. The use of metal foam projectiles to
simulate shock loading on a structure. International Journal of Impact Engineering (in
press).
Radford, D. D., Fleck, N.A., Deshpande, V.S., 2005b. The response of clamped sandwich
beams subjected to shock loading. International Journal of Impact Engineering (in press).
Rathbun, H. J., Radford, D.D., Xue, Z., Yang, J., Deshpande, V.S., Fleck, N.A.,
Hutchinson, J.W., Zok, F.W., Evans, A.G., 2004. A dynamic probe for validating
simulations of shock loaded metallic sandwich panels. International Journal of Solids and
Structures (submitted).
Stout, M. G. Follansbee, P.S., 1986. Strain rate sensitivity, strain hardening, and yield
behavior of 304L stainless steel. Trans. ASME: Journal of Engineering Materials
Technology 108: 344-353.
Symmonds, P. S., 1954. Large plastic deformations of beams under blast type loading.
Second US National Congress of Applied Mechanics.
Wang, A. J., Hopkins, H.G., 1954. On the plastic deformation of built-in circular plates
under impulsive load. Journal of Mechanics and Physics of Solids 3: 22-37.
16
Xue, Z., Hutchinson, J.W., 2003. Preliminary assessment of sandwich plates subject to
blast loads. International Journal of Mechanical Sciences 45: 687-705.
Xue, Z., Hutchinson, J.W., 2004. A comparative study of blast-resistant metal sandwich
plates. International Journal of Impact Engineering 30: 1283-1305.
17
Specimen
o p o oI l vρ=
(kNsm-2)
pρ
(kgm-3)
0l
(mm)
0v
(ms-1)
Mid-span
deflection of back
face (mm)
Cε
M1 3.26 416 49 160 0.6 -
M2 3.69 371 49 205 1.3 -
M3 4.59 505 50 183 2.2 -
M4 6.12 457 49 271 4.0 -
M5 8.11 480 48 352 7.2 -
M6 8.97 457 49 400 8.2 -
M7 10.51 490 47 456 10.6 -
M8 11.75 493 48 496 12.2 -
M9 13.07 493 48 552 14.2 -
A1 3.26 433 47 160 2.0 0.07
A2 6.18 388 50 321 4.4 0.39
A3 9.93 509 45 434 8.1 0.71
A4 13.31 503 48 551 12.8 0.78
B1 3.15 393 51 157 1.1 0.05
B2 6.19 487 48 265 3.1 0.21
B3 9.54 490 47 414 5.7 0.48
B4 12.87 470 48 570 9.2* 0.81
Table 1: Summary of the dynamic experiments performed on circular monolithic and
sandwich plates. The specimens labelled M denote monolithic plates, A denotes sandwich
plates of core thickness c = 10 mm, and B denotes sandwich plates of core thickness
c = 22 mm. The superscript * indicates that tensile failure of the front face occurred.
18
List of Figures
Figure 1. Sandwich plate geometry.
Figure 2. Sketch of the clamping arrangement. All dimensions are in mm.
Figure 3. (a) Quasi-static compressive response for foam specimens of side-length
L = 10 mm and 22 mm in the sandwich configuration; (b) Quasi-static tensile response of
the AISI 304 stainless steel; results are shown for each sheet thickness, h.
Figure 4. (a) The dynamic strength enhancement ratio R as a function of plastic strain rate pε& for 304 stainless steel (Stout and Follansbee 1986). (b) Estimated tensile stress versus
strain histories for 304 stainless steel (h = 1.18 mm) at three selected values of the applied
strain rate.
Figure 5. Measured permanent back-face deflection at mid-span of the dynamically
loaded monolithic and sandwich plates, as a function of the initial foam projectile
momentum I0.
Figure 6. Measured permanent core compression cε at the mid-span of the dynamically
loaded sandwich plates, as a function of the initial foam projectile momentum I0. Finite
element (FE) predictions (rate dependent model) for cε are included.
Figure 7. Photographs of the dynamically tested specimens (a) M9, (b) A4 and (c) B4. All
these plates were tested at =0I 13 kNs m-2 and sectioned along their diametral plane.
Figure 8. Photographs of the quasi-statically tested (a) monolithic plate (b) 10c = mm
sandwich plate and (c) 22c = mm sandwich plate. These plates were loaded until the
permanent back-face deflection at the mid-span matched that obtained in the
corresponding dynamic experiments of Fig. 7 and sectioned along their diametral plane.
19
Figure 9. Finite element predictions of the mid-span deflection versus time histories of
monolithic plate specimen M9 and 22c = mm sandwich plate specimen B4. The figure
shows the mid-span deflection of the back-face of the monolithic plate, and of both faces
of the sandwich plate. =0I 13 kNs m-2.
Figure 10. Measured and predicted back-face deflections for the monolithic plates.
Figure 11. Measured and predicted back-face deflections for the (a) 10c = mm and
(b) 22c = mm sandwich plates.
Figure 12. The predicted pressure versus time histories exerted by the foam projectiles on
the monolithic specimen M9, and the sandwich plate specimens A4 ( 10c = mm) and B4
( 22c = mm). =0I 13 kNs m-2.
Figure 13. Predicted back-face deflections at mid-span. (a) Sandwich plate ( 10c = mm)
and monolithic plate loaded by projectile, and monolithic plate loaded by pressure
transient; (b) Sandwich plate ( 22c = mm) and monolithic plate loaded by projectile, and
monolithic plate loaded by pressure transient.
20
Figure 1: Sandwich plate geometry.
c
h
h
R = 80 mm
face sheets
Alporas Foam Core
d/2 = 14.3 mm
v0
metallic foam projectile
L C
21
Figure 2: Sketch of the clamping arrangement. All dimensions are in mm.
L
specimen
frame
clamping ring
A A
c
frame
specimen
clamping rings 7
350 R = 80
R = 125
Section A-A
25
Plan View
22
Figure 3. (a) Quasi-static compressive response for foam specimens of side-length L = 10 mm and 22 mm in the sandwich configuration; (b) Quasi-static tensile response of the AISI 304 stainless steel; results are shown for each sheet thickness, h.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90
25
50
75
100
c = 22 mm
nom
inal
stre
ss (M
Pa)
nominal strain
c = 10 mm
0.0 0.1 0.2 0.3 0.4 0.5 0.60
250
500
750
1000
1250
true
stre
ss (M
Pa)
logarithmic strain
h = 0.88 mm
h = 1.18 mm
h = 2.92 mm
(a)
(b)
23
Figure 4. (a) The dynamic strength enhancement ratio R as a function of plastic strain rate pε& for 304 stainless steel (Stout and Follansbee 1986). (b) Estimated tensile stress versus
strain histories for 304 stainless steel (h = 1.18 mm) at three selected values of the applied strain rate.
1.0
1.2
1.4
1.6
1.8
1031
R
plastic strain rate εp (s-1)
ε p = 0.1
.10-3
0.0 0.1 0.2 0.3 0.4 0.5 0.60
500
1000
1500
2000
110-3
.
104
102
true
stre
ss (M
Pa)
logarithmic plastic strain
εp (s-1)
(a)
(b)
24
0 2 4 6 8 10 12 140
5
10
15
sandwich plate(c = 22 mm)
m
id-s
pan
back
face
def
lect
ion
(mm
)
I0 = ρp l0v0 (kNs m-2)
monolithic plate
sandwich plate (c = 10 mm)
Figure 5. Measured permanent back-face deflection at mid-span of the dynamically loaded monolithic and sandwich plates, as a function of the initial foam projectile momentum I0.
25
0 2 4 6 8 10 12 140.0
0.2
0.4
0.6
0.8
1.0
c = 22 mm (experiment)
I0 = ρp l0v0 (kNs m-2)
εc
c = 22 mm (FE)
c = 10 mm (FE)
c =10 mm (experiment)
Figure 6. Measured permanent core compression cε at the mid-span of the dynamically loaded sandwich plates, as a function of the initial foam projectile momentum I0. Finite element (FE) predictions (rate dependent model) for cε are included.
26
Figure 7. Photographs of the dynamically tested specimens (a) M9, (b) A4 and (c) B4. All these plates were tested at =0I 13 kNs m-2 and sectioned along their diametral plane.
(a)
(b)
(c)
27
Figure 8. Photographs of the quasi-statically tested (a) monolithic plate (b) 10c = mm sandwich plate and (c) 22c = mm sandwich plate. These plates were loaded until the permanent back-face deflection at the mid-span matched that obtained in the corresponding dynamic experiments of Fig. 7 and sectioned along their diametral plane.
(a)
(b)
(c)
28
0.0 0.5 1.0 1.5 2.00
5
10
15
20
25
30
m
id-s
pan
defle
ctio
n (m
m)
time (ms)
front face of sandwich plate
back face of monolithic plate
back face of sandwich plate
Figure 9. Finite element predictions of the mid-span deflection versus time histories of monolithic plate specimen M9 and 22c = mm sandwich plate specimen B4. The figure shows the mid-span deflection of the back-face of the monolithic plate, and of both faces of the sandwich plate. =0I 13 kNs m-2.
29
0 2 4 6 8 10 12 140
5
10
15
20
mid
-spa
n ba
ck fa
ce d
efle
ctio
n (m
m)
I0 = ρp l0v0 (kNs m-2)
FE (rate independent)
FE (rate dependent)
measurement
Figure 10: Measured and predicted back-face deflections for the monolithic plates.
30
Figure 11. Measured and predicted back-face deflections for the (a) 10c = mm and (b) 22c = mm sandwich plates.
0 2 4 6 8 10 12 140
5
10
15
20
I0 = ρp l0v0 (kNs m-2)
FE (rate dependent)
mid
-spa
n ba
ck fa
ce d
efle
ctio
n (m
m)
measurement
FE (rate independent)
0 2 4 6 8 10 12 140
5
10
15
FE (rate dependent)
mid
-spa
n ba
ck fa
ce d
efle
ctio
n (m
m)
I0 = ρp l0v0 (kNs m-2)
measurement
FE (rate independent)
(a)
(b)
31
Figure 12. The predicted pressure versus time histories exerted by the foam projectiles on the monolithic specimen M9, and the sandwich plate specimens A4 ( 10c = mm) and B4 ( 22c = mm). =0I 13 kNs m-2.
0.00 0.05 0.10 0.15 0.200
50
100
150
200
monolithic plate sandwich plate, c = 10 mm sandwich plate, c = 22 mm
pr
essu
re (M
Pa)
time (ms)
32
Figure 13: Predicted back-face deflections at mid-span. (a) Sandwich plate (c = 10mm) and monolithic plate loaded by projectile, and monolithic plate loaded by pressure transient. (b) Sandwich plate (c = 22mm) and monolithic plate loaded by projectile, and monolithic plate loaded by pressure transient.
0 2 4 6 8 10 12 140
5
10
15
20
mid
-spa
n ba
ck fa
ce d
efle
ctio
n (m
m)
I0 = ρp l0v0 (kNs m-2)
monolithic plate (projectile loading)
monolithic plate (pressure loading)
sandwich plate(projectile loading)
0 2 4 6 8 10 12 140
5
10
15
20
monolithic plate (pressure loading)
monolithic plate (projectile loading)
mid
-spa
n ba
ck fa
ce d
efle
ctio
n (m
m)
I0 = ρp l0v0 (kNs m-2)
sandwich plate(projectile loading)
(a)
(b)